- T-0201
Neural networks lack compositional generalization. They cannot reliably combine known concepts in novel ways, a core requirement for general reasoning.
Survival 67%
Challenges
AGI requires combining neural networks with symbolic reasoning—explicit knowledge representation, formal logic, and structured world models that neural nets alone cannot provide.
3 claims · 2 surviving · 0 contested
Claims
- T-0202
World models require structured representation. Implicit knowledge in weights is insufficient for planning, counterfactual reasoning, and causal inference.
- T-0203
Neuro-symbolic integration is tractable. Differentiable programming, neural theorem provers, and learned symbolic abstractions show viable paths to hybrid systems.